Automated Gait Recognition

Abstract

Biometric technologies possess considerable importance in the modern world where real time and remote surveillance for identification of individuals is critical. Biometric identifiers may depend on physiological or behavioral characteristics. For the scope of this project, we aim to focus on gait recognition that serves as a behavioral identifier. Gait recognition involves recognizing people by the way they walk. It possesses an edge over existing biometric technologies since it is able to identify individuals without their cooperation - a factor that makes it suitable for multiple use cases where a user may not volunteer to identify himself. Existing Biometrics can be easily disguised such as fingerprint can be forged; face can be faked using masks but gait is the only biometric which is very difficult to disguise. We aim to determine the gait signature of an individual from a sequence of images/ videos and perform human recognition through comparison with the gait signature.